Page 1
Page 1
Elevating video content creation with ai assistance = ارتقاء إنشاء محتوى الفيديو بمساعدة الذكاء الاصطناعي
We developed an AI Assistant equipped with features such as description crafting, title generation, keyword extraction, image captioning, clickbait detection, and sentiment analysis.To achieve these functionalities, we proposed a model for generating video descriptions using ResNet50 as a feature extractor and a LSTM network with an attention mechanism as a sequence generator, achieving a BLEU-1 score of 0.907 and a ROUGE-L score of 0.645. For keyword extraction, we utilized Sentence Transformer to identify strategically relevant keywords from the generated descriptions. For title generation, we fine-tuned the BART model, achieving a ROUGE-L score of 0.45. For clickbait detection, we used SVC classifier with linear kernel and TF-IDF vectorization for feature extraction, resulting in 96% accuracy. Our sentiment analysis model using a CNN-LSTM architecture achieved 80% accuracy in analyzing comments on videos. For image captioning, we employed a feature extractor with a CNN layer followed by an LSTM model, achieving a BLEU-1 score of 0.53. Our platform empowers creators by simplifying complex tasks and offering deeper audience engagement insights, making it a powerful tool in the evolving digital content creation.
Egocentric video summarization
Video summarization is defined as the generation of a summary of extensive video content that comes from all kinds of videos including egocentric videos by detecting and presenting the material to potential users which is most informative and contains interesting information. Video summarization has many practical applications and Egocentric video summarization approaches have been proposed to solve various problems in the healthcare industry. This work focuses on Alzheimer. Patients suffering from Alzheimer’s face difficulties in remembering what happened during their day, the identity of persons and medicine they took.
